> For the complete documentation index, see [llms.txt](https://help.multiomics.illumina.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.multiomics.illumina.com/icm/analyses/walkthroughs/correlation-engine-in-connected-multiomics.md).

# Correlation Engine in Connected Multiomics

This walkthrough starts with a list of genes in Connected Multiomics and uses the integrated Correlation Engine capabilities to turn it into biological meaning with supporting evidence.

This walkthrough uses Illumina StrataMap Mouse Kidney data from one sample. It is available in the Multiomics demo data bundle.

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-43a72c932e1c858c0f539d4f936dd6e4f34e74ce%2Fuse%20the%20mouse%20kidney%20data.png?alt=media" alt=""><figcaption><p>Illumina StrataMap Mouse Kidney data can be found in the Multiomics Demo Data bundle</p></figcaption></figure>

## 1. Start with an interesting signature

Let’s imagine we have completed our [Differential analysis](https://help.multiomics.illumina.com/icm/analyses/analysis-functionality/task-menu/statistics/differential-analysis.md) comparison. We identified a set of genes that differ between the conditions we are studying and we have filtered to the genes of interest (e.g. by significance and fold change).

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-0dc3cdebab07f2c916c8e4457f109bf7e06836f6%2Fuse%20a%20filtered%20DE%20analysis%20node.png?alt=media" alt=""><figcaption><p>Start with a list of interesting features</p></figcaption></figure>

A volcano plot, heatmap, or gene list tells us which genes changed, but our next question is usually much more biological: **What does this pattern actually mean?**

Rather than exporting results for a separate literature or database search, we can interrogate the signature directly in Connected Multiomics using Correlation Engine.

* The filtered feature result for Podocytes versus Principal cells contains 652 genes. Click the icon to view Correlation Engine annotations for each gene.

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-dcff72361371ac83c51fad9b0f729e02ee99e7e8%2Fatlases%20report%201.png?alt=media" alt=""><figcaption><p>The filtered feature result shows 652 interesting genes. View Correlation Engine annotations for each gene by clicking the icon.</p></figcaption></figure>

## 2. [Correlation Engine pathways](https://help.multiomics.illumina.com/icm/analyses/analysis-functionality/task-menu/biological-interpretation/correlation-engine-pathways.md)

The first question we might ask is: **what biological processes are represented by these changes?**

Correlation Engine Pathways lets us take this molecular signature and evaluate it against curated biological gene sets. Rather than treating this simply as a list of individual genes, we are looking for biological programs that are represented by the overall signature.

* Click the Filtered feature list result node, open **Biological interpretation → Correlation Engine pathways** task from the context sensitive menu

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-cc9e49d778b3f75f5780941600dc58af773e8ece%2FCorrelation%20engine%20pathways.png?alt=media" alt=""><figcaption><p>Click the Correlation Engine pathways task in the menu under Biological interpretation to start the task.</p></figcaption></figure>

* After the task completes, double-click the Correlation Engine pathways results node to open the report.

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-ee9bbde4ece33241ebe5b5272a47b549eefae94d%2FCorrealtion%20engine%20pathways%20report.png?alt=media" alt=""><figcaption><p>Sort, filter, download, and explore the report. Open individual gene sets or view eligible results in Data Viewer.</p></figcaption></figure>

{% hint style="info" %}
You can visualize pathway results in Data Viewer only when the report has fewer than 100 pathways. The automatic Data Viewer session includes the top 30 pathways by normalized enrichment score.
{% endhint %}

The first result is gene set [M1533](https://www.gsea-msigdb.org/gsea/msigdb/cards/CUI_TCF21_TARGETS_2_DN): significantly down-regulated genes in kidney glomeruli from TCF21 \[Gene ID=6943] knockout mice. This top result already points to kidney-related biology.

* Filter to results for more focused processes related to the kidney.

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-aa09d727a622a3e6de52e1cd2e15247a28b7c057%2FCorrelation%20engine%20pathways%20report%20filtered.png?alt=media" alt=""><figcaption><p>Customize the report</p></figcaption></figure>

## 3. [Correlation Engine atlases](https://help.multiomics.illumina.com/icm/analyses/analysis-functionality/task-menu/biological-interpretation/correlation-engine-atlases.md)

Pathways tell us *what biological processes may be involved*. Sometimes we need more context: **Where in biology would we expect to see a signature like this?**

Instead of only asking which pathway is enriched, the Atlases let us examine the signature through different biological lenses and understand the contexts associated with the signature.

* In the context-sensitive menu, select **Biological interpretation → Correlation Engine atlases**.

The output includes **Body, Disease, Knockdown, and Pharmaco** atlas reports for interpreting differential-analysis results.

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-b718cf2d7bf86e5b6bfaee171d949c4bdf83f1d5%2FCorrelation%20engine%20atlases.png?alt=media" alt=""><figcaption><p>Correlation Engine atlases includes <strong>Body, Disease, Knockdown, and Pharmaco</strong> atlas reports for interpreting differential-analysis results.</p></figcaption></figure>

**Disease Atlas:** What diseases or disease-related contexts resemble my signature?

**Body Atlas:** What tissue or biological contexts are associated with this pattern?

**Knockdown Atlas:** What gene perturbations produce related molecular responses?

**Pharmaco Atlas:** What pharmacological contexts are associated with this signature?

* Double-click to open the Body atlas report.
* Switch to the **Single cell tab**
* Open **Configure** > **Content** > Sort by **Direction** > Order as **Descending**

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-052e2eabd8f90958cc3d24a03ef6e8672ecfba3c%2Fimage%20(621).png?alt=media" alt=""><figcaption><p>Explore and sort the results</p></figcaption></figure>

The top body system is the urogenital system, with corresponding cell types. This fits the kidney dataset because the kidney belongs to the broader urogenital system. It shows that Correlation Engine recovers biological context consistent with the analyzed tissue.

The single cell Body Atlas results for querying the ‘Podocyte vs Principal cell’ signature was significantly correlated to a wide array of over 3,000 independent curated single cell signatures. In the top results, 4 up-regulated Podocyte (positive correlations) and 5 down-regulated Principal cell (negative correlations) validate the ICM results which could be explored more deeply in Correlation Engine.

## 4. [Correlation Engine meta-analysis](https://help.multiomics.illumina.com/icm/analyses/analysis-functionality/task-menu/biological-interpretation/correlation-engine-meta-analysis.md)

Now suppose we have identified an interesting biological signal. The next question might be: **Has this pattern been observed elsewhere?**

Meta-analysis lets us compare our molecular signature against independently curated research in Correlation Engine. Instead of stopping at the observation in our experiment, we can look for supporting evidence across other studies and biological contexts.

The goal isn't simply to find the smallest p-value. We are asking whether independent biological evidence supports the pattern we are observing.

* Open the Correlation Engine meta-analysis report
* Click the **Common genes** column title twice to sort the column

The result is designed to include Study/Bioset information along with scores, specificity, direction, P-values, and common-gene information, with the ability to continue exploring results in Data Viewer or follow links into Correlation Engine.

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-226376fe12f9e952bb4f68107af66eaa0662fd39%2Fmeta%20analysis%20report.png?alt=media" alt=""><figcaption><p>The Correlation Engine meta analysis report includes Bioset ID and Study ID links directly to Correlation Engine</p></figcaption></figure>

Correlation Engine is comparing the molecular signature against previously characterized Biosets. Reassuringly, we see strong kidney and glomerular results, which are biologically relevant to our podocyte comparison. But we also see signatures from other tissues, giving us an opportunity to investigate biological programs that may be shared across tissue contexts. We can also ask: **which genes are actually driving the relationship?**

## 5. [Correlation Engine bioset correlation](https://help.multiomics.illumina.com/icm/analyses/analysis-functionality/task-menu/biological-interpretation/correlation-engine-bioset-correlation.md)

Meta-analysis helps us understand the broader evidence. But sometimes we want to know something more specific: **Which individual experimental signatures closely resemble mine?**

Correlation Engine meta-analysis asks whether the signature replicates across studies, while bioset correlation asks which specific studies show the strongest pairwise concordance. Bioset correlation compares our signature against individual BioSets in Correlation Engine and helps identify specific studies showing concordant or discordant molecular patterns.

* Open the Correlation Engine bioset correlation report.

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-0da5fd2fe6827459e5c3e5f65ce1356f1045d590%2Fbioset%20correlation%20report.png?alt=media" alt=""><figcaption><p>Correlation Engine bioset correlation report</p></figcaption></figure>

Remember here we are asking: **Which specific experimental signatures most closely resemble or oppose my signature?**

Among the strongest results, we immediately recover several kidney-associated Biosets. One particularly relevant example comes from podocytes in a diabetic mouse model. We have 231 genes in common and a highly significant association. Notice that the correlation direction is negative. That tells us the relationship is inverse, which gives us another dimension to investigate rather than simply asking whether two experiments contain overlapping genes. We can also use View Top Features to investigate which genes are contributing to an interesting relationship.

* Filter the Bioset result by 'podocyte'

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-d992bf8eb74a22844ae3f283cb1c0a681ea3d12c%2Fimage%20(622).png?alt=media" alt=""><figcaption></figcaption></figure>

This filters to other podocyte experiments spanning different diseases and treatments, where the negatively correlated genes within each could provide key mechanistic insight.

## 6. [Ask Correlation Engine](https://help.multiomics.illumina.com/icm/analyses/analysis-functionality/task-menu/ask-correlation-engine.md)

So far we've been using structured analytical tasks. But sometimes we simply have a biological question and want to explore biological evidence in public studies and datasets.

Ask Correlation Engine gives us a way to ask a biological question directly to receive answers grounded in evidence retrieved from Correlation Engine. Examples might be: **In which tissues is this gene expressed? What diseases are associated with this gene?**

WT1 is one of the genes we're interested in, so I can simply ask: **In which cell types is WT1 expression highest?**

<figure><img src="https://580316046-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FWMxqQAMFOJtu98OBk9KN%2Fuploads%2Fgit-blob-1c63af0520e74b609c43020fe5cc08a8900a4376%2Fimage%20(625).png?alt=media" alt=""><figcaption><p>Ask Correlation Engine (found at the bottom of the task graph) also links out to explore in Correlation Engine</p></figcaption></figure>

Correlation Engine returns an evidence-based answer from its available data. In the single-cell panel, podocytes of the kidney are the top-ranked cell type for WT1 expression, followed by two additional kidney cell populations.

The response provides supporting evidence and describes its limitations. This context helps interpret the result. Select the button to explore further in Correlation Engine.

## Bring the story together

Across the Correlation Engine analyses, the podocyte versus principal cell signature consistently recovered kidney and podocyte relevant biology by connecting the signature to related experimental evidence.

Notice what happened throughout this workflow.

1. We started with a statistical result: **these genes changed.**
2. Pathway analysis helped us ask: **What biology is represented?**
3. Atlases helped us ask: **In what biological contexts do we see something similar?**
4. Meta-analysis helped us ask: **Is there supporting evidence across independent studies?**
5. Bioset correlation helped us ask: **Which specific experimental signatures resemble ours?**
6. Ask Correlation Engine gave us a way to explore additional biological questions.

| Step                                  | Question                                                 |
| ------------------------------------- | -------------------------------------------------------- |
| Differential analysis                 | What features changed in my comparison?                  |
| Correlation Engine Pathways           | What biological processes are represented?               |
| Correlation Engine Atlases            | Where else in biology do we see this pattern?            |
| Correlation Engine Meta-analysis      | Is there supporting evidence across independent studies? |
| Correlation Engine Bioset Correlation | Which specific experimental signatures resemble mine?    |
| Ask Correlation Engine                | What else can I learn about this finding?                |

\
The value is not simply generating a plot. It is moving from a molecular signal to supporting evidence and the next research question.

{% hint style="info" %}
To export differentially expressed genes or proteins from Connected Multiomics and upload them into Correlation Engine for deeper biological interpretation please see the next [walkthrough](https://help.multiomics.illumina.com/icm/analyses/walkthroughs/correlation-engine-for-gene-discovery.md) which leads users through Connected Multiomics export > Correlation Engine Import. The new functionality as shown here, in the previous sections, does not require users to do this, as a new Study gets created in Correlation Engine-standalone from Connected Multiomics.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://help.multiomics.illumina.com/icm/analyses/walkthroughs/correlation-engine-in-connected-multiomics.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
